Chromosome Behaviour at Meiotic Cell Divisions

Maj Hultén
Department of Biological Sciences, University of Warwick, Coventry, UK

Derived from the Greek word for diminution, meiosis means “to halve”. The basic purpose of meiosis is straightforward, i.e. to reduce by half the chromosome complement of a diploid cell during gametogenesis to give rise to haploid gametes. Subsequent fusion of a haploid sperm cell and a haploid oocyte at fertilization restores the somatic diploid chromosome number in the resulting offspring.

The process of meiosis is conserved through evolution from yeast to man although with marked differences between species and also between sexes. Unlike in mitosis where, in principal, daughter cells should have the same genetic component as the parent cell from which they were derived, meiosis serves to effectively “shuffle” the genetic material in two ways. These shuffling processes generate an almost infinite genetic variety from the starting parental chromosomes.

The first of the mechanisms giving rise to variation in alleles is by the independent segregation of the chromosome pairs to the resulting cells; in humans this mechanism alone would give rise to 223, i.e. 8,388,608 different genetic cell types, and is termed ‘random assortment’. The second mechanism occurs when chromosome arms exchange homologous regions of DNA through crossing-over and the formation of chiasmata, leading to recombination of parental alleles.

It is exceedingly important to note that chiasma formation, crossing-over and reciprocal recombination are three different ways of describing the very same event. As will be described in some detail in the following sections, chiasma/crossing-over/recombination positions along the length of individual chromosomes can be pinpointed by cytogenetics. On the other hand, tracking DNA markers between parents and children allows identification of the recombination of the respective alleles.

There are two meiotic divisions preceded by a single round of DNA replication

Following premeiotic DNA replication cells enter meiosis with each chromosome being composed of two chromatids, held together at the centromere. The meiotic process per se involves two cell divisions, and is accordingly divided into two parts, meiosis I and meiosis II (Fig 1). As in mitosis these divisions are sub-divided into the stages Prophase, Metaphase, Anaphase and Telophase, thus a full meiotic process includes two of each of these stages. A prolonged Prophase in the first meiotic division (PI) is involved in preparation for the halving of chromosome number, this “reductional” division taking place at Anaphase I (AI). The salient points of each stage are highlighted in Table 1.

Homologue pairing and crossing-over/chiasma formation takes place at PI

The most complex part of meiosis concerns the prolonged Prophase (PI) taking place at the first meiotic “reductional” division in order to half the chromosome number at AI. Most importantly, the preparation for the MI/AI spindle to carry its cargo of whole chromosomes (rather than chromatids as in mitosis) involves intimate pairing and association of all four chromatids of the maternal and paternal homologues. This intimate pairing (synapsis) is mediated by a proteinaceous structure, called the Synaptonemal Complex (SC). Breakage and reunion (crossing-over) between non-sister chromatids gives rise to configurations known as chiasmata. As already stressed above, crossing-over points corresponding to the sites where chiasmata are formed can be visualised by cytogenetics at the PI stage, this then using immuno­fluorescence with a DNA mismatch protein called MLH1.

The chiasmata hold homologous chromosomes together locally during the next phase, when they separate at Diplotene. It is at this later stage that crossing-over points can first be seen as chiasmata by miscroscopy, using traditional cytogenetic preparations and staining methods.

Most studies on chiasmata has been performed at the Diakinesis/MI stages

Following chromatin condensation during the intermediate Diakinesis stage chromosome pairs (bivalents) align on the MI plate. Most information on the frequency and distribution of chiasmata along the length of individual chromosomes has been obtained by analysis of spermatocytes at the Diakinesis/MI stages, using conventional air-dried preparations of testicular biopsy samples with slides stained by Q-banding, Orcein and C-banding.

Figure 1: Schematic illustration of meiosis with (a) homologous chromosome synapsis, and crossing-over at the pachytene stage of prophase I and the derivative bivalents at MI, and (b) progression through metaphase I to anaphase I, metaphase II to anaphase II, and telophase II showing the four potential haploid gametes.

Figure 2: Schematic illustration of the relationship of a crossover, as viewed directly (as a chiasma or MLH1 focus) and the meiotic products (gametes). There are four chromatids, each a potential gamete, only two of which are recombinant.

Figure 3: Schematic illustration of the SC.

Table 1. Salient points of meiotic stages

Stage Sub-division Features
Prophase I   DNA replication of each chromosome to give sister chromatids held together at the centromere
  Leptotene Start of chromosome condensation, formation of the axial elements of the synaptonemal complex
  Zygotene Chromosome pairing; completion of the formation of the synaptonemal complex
  Pachytene Chromosomes fully paired, cross-overs/chiasmata fully established
  Diplotene Initiation of separation of the synapsed chromosomes
  Diakinesis Chromosomes condense and become fully separated, except at points of crossing-over/chiasma formation.
Metaphase I   Homologous chromosomes align on the equatorial plate with the kinetochores being the point of attachment to the spindles
Anaphase I   Reductional division; homologous pairs separate, but sister chromatids remain together
Telophase I   Formation of two daughter cells with the haploid chromosome number
Prophase II   Nuclear envelope dissolves; creation of a new spindle
Metaphase II   Chromosomes align on the spindle
Anaphase II   Separation of centromeres; migration of sister chromatids to opposite poles
Telophase II   Further cell division resulting in four potential haploid gametes from each parent cell

The bivalents are then much shorter than at the previous Diplotene stage, which makes it easier to separate them from each other. One drawback here is the relative scarcity of spermatocytes at the Diakinesis/MI stage in these slides; and it has not yet been possible to obtain the same type of information on oocytes, where Diakinesis/MI takes place in individual cells just before ovulation (see further below).

The subsequent AI and TI cell divisions are expected to give rise to two daughter cells, containing the haploid chromosome number.

Meiosis II is similar to mitosis

After a brief Prophase II (PII) stage, taking place without preceding DNA replication, the two daughter cells complete meiosis II to produce four end products. This “equational” division is similar to mitosis. It should be recognized though that, following crossing-over/chiasma formation, normally at least one of the sister chromatids (potential gametes) of each chromosome now is recombinant, containing a combination of variant genes (alleles) from the maternal and paternal homologues (Fig 2).

There are drastic variations in the meiotic process between males and females

There are drastic differences in behaviour between human male and female germ cells during gametogenesis and meiosis. This includes in particular the timing and continuity of events, the pairing and crossover processes, chromosome segregation and especially the actual gamete produced – two cells as diverse as the oocyte and the sperm could hardly be imagined to derive from the same process! These differences are summarised in Table 2.

Meiosis in males starts after puberty and continues throughout life; and normally the production is at least 60 millions sperm daily. Meiosis in females, on the other hand, begins around the 12th week of fetal life with homolog pairing, crossing-over and chiasma formation up until around week 20. Oocytes then arrest at the diplotene stage and are generally thought to undergo extensive cell death with numbers dropping from around 7 million to around 2 million at birth.

Female meiosis does not resume until puberty, when after the mid cycle Luteinising Hormone surge, chromosomes line up on the MI plate and undergo the reductional division at AI just before ovulation, most commonly of a single oocyte. This division is asymmetrical, where most of the cytoplasm is retained by one of the daughter cells, which will form the future mature oocyte. The other daughter cell forms a small polar body, which soon degenerates. The number of oocytes ovulated are estimated to be less than 500 during the life time of women.

Following progression through the second prophase, oocytes yet again arrest at MII until fertilisation occurs. At fertilization, once the sperm has entered the oocyte and caused activation, oocyte meiosis continues through AII and TII, resulting in the ejection of the small second polar body but retention of the oocyte.

Table 2.  Summary of the main differences between human male and female meiosis

Male

Female

Meiosis is a continuous process from puberty throughout life, and can result in the production of around 200-300 million spermatozoa daily. A full cycle of spermatogenesis takes approximately 120 days, of which 72-74 days are spent during meiosis

Meiosis can take over 40 years from start to finish and only a few oocytes actually progress to the final stages, most being lost before birth

Each parent cell produces 4 gametes (spermatozoa), all divisions being equal

The two cell divisions are unequal with most cytoplasm retained in the oocyte and only a minor part forming the first and second polar bodies, thus only one actual gamete is produced from each parent cell

The testis contains a population of stem cells which give rise to the continuing supply of gametes

Oocyte numbers appear to be limited to those present at birth with around 350 ovulating between puberty and the menopause. However in mice, recent experiments have suggested the existence of stem cells that may constitute a reserve.

Chromosome synapsis is very efficient, initated near the ends of chromosomes in karyotypically normal, fertile men

Chromosome synapsis is less efficient and interstitial initiation  more common in females

Chiasma formation is an efficient process

Chiasma formation is a less efficient process

Lower overall chiasma frequency

Higher overall chiasma frequency

Synaptonemal complexes are condensed

Synaptonemal complexes are relatively decondensed, having almost twice the total length per cell compared to in males.

Tendency of chiasmata to occupy preferential positions with hotspots near the ends of chromosomes

Tendency of chiasmata to occupy preferential positions slightly more interstitially

Homologue pairing is facilitated by a meiosis-specific protein structure - the SC

Meiotic chromosome pairing at PI involves three successive developmental stages, homologue recognition, presynaptic alignment and intimate synapsis. To date nothing is known of the long distance homolog recognition. During the initial development of the Synaptonemal Complex (SC) most of the DNA forms large loops emanating from the sides of the SC in a “chromatin cloud” (Fig 3). The SC is a protein lattice that resembles railroad tracks and connects paired homologous chromosomes in most meiotic systems. The two side rails of the SC, known as lateral elements, are connected by proteins known as transverse filaments. Thus the SC forms a protein scaffold for meiotic chromosome pairing and crossing-over/chiasma formation/recombination during the PI substages Leptotene (when the so-called axial elements are formed but no actual synapsis is taking place), Zygotene (when synapsis starts and the axial elements are transformed to the lateral elements of the SC) and Pachytene (when synapsis is completed and the SC consists of lateral elements, transverse filaments and a central element). The SC then breaks down at the following Diplotene stage, which is intermediate between PI and Diakinesis/MI.

Chiasma formation is visualised at Pachytene by MLH1 immunofluorescence

Intimate homologue synapsis is mediated by special proteins such as RAD51 and DMC1. The subsequent crossing-over and chiasma formation involves a number of DNA mismatch repair proteins, including in particular MLH1 (bacterial MutL homologue 1). The application of anti-MLH1 to SC preparations at the pachytene stage shows a labelling pattern consisting of distinct foci, always precisely associated with the SC and never in closely juxtaposed pairs (Fig 4). To date more information on the frequency and distribution of crossing-over points/chiasma formation/recombination along the length of individual chromosomes has been achieved on spermatocytes by analysis of SC preparations made from testicular biopsies of adult men than on oocytes obtained from fetal ovarian biopsies.

The frequency and distribution of MLH1 foci agrees with that of chiasmata at Diakinesis/MI of spermatocytes (Fig 5); and there is now no doubt that MLH1 is an appropriate marker for chiasma formation. The average cross-over/chiasma/ recombination frequency estimated by MLH1 analysis in human males is around 50 with a range of 40-60, while that in females is around 70 but with a much larger variation between individuals than in males. Recent SC and MLH1 observations on human fetal oocytes at the pachytene stage during fetal development also quite clearly demonstrate that failure of synapsis and chiasma formation is much more common than at the corresponding stage in males.

There is a large database available on chiasma patterns in human males

A large database of patterns of chiasmata has accumulated over the last three decades by the study of Diakinesis / MI spermatocytes in preparations from testicular biopsy samples. This has demonstrated that chiasma formation is a dynamic process, varying somewhat between chromosomes,  cells and  subjects.  Thus we have quite good knowledge on the basic patterns and its variation between individual chromosomes, cells, and subjects in normal fertile men, as well as changes associated with infertility, and in carriers of structural chromosome rearrangements.

Figure 4: MLH1 foci on SCs of spermatocyte (a) and oocyte (b). Note the near-telomere preference in the spermatocyte and the higher number of foci in the oocyte.

Figure 5: The morphology of MI and MII chromosomes is very different.

There is surprisingly little variation in the pattern of chiasma frequency and distribution of chiasmata along the lengths of individual chromosomes in normally fertile human males.

The average cross-over/recombination frequency estimated from chiasma analysis at the Diakinesis/MI in human males agrees with that from MLH1 analysis of SC preparations at the PI stage, and is thus around 50 with a range of 40-60. It is important to recognize that so far the same type of information on chiasmata at the Diakinesis/MI stage in human females is still lacking. Female Diakinesis/MI takes place just before ovulation usually in single oocytes; and the main reason for this lack of knowledge is the difficulty in obtaining enough material of oocytes for study.

In spite of quite detailed knowledge on the patterns of crossing-over/chiasmata/recombination obtained by cytogenetic studies, our knowledge on the underlying mechanisms of regulation of the basic pattern is still very poor indeed.

At least one chiasma is required to secure proper Anaphase I segregation

One important role of chiasma/crossover formation is to link chromosome pairs in such a way that whole chromosomes are conveniently transported to daughter cells at AI. In the normally fertile human male at least one chiasma/crossover is generally formed per chromosome pair (bivalent) irrespective of its length; this chiasma/crossover is considered to be obligate to secure “regular” segregation of the chromosomes involved, by allowing the maternal and paternal chromosomes of the bivalent to be properly orientated on the MI spindle. In meiosis mono-orientation of centromeres/kinetochores is required to ensure that at the first division chromosomes segregate, rather than chromatids which segregate at the second division. As already stressed, there is currently no direct information on patterns of chiasmata at the MI stage in human oocytes but recent SC and MLH1 observations on human fetal oocytes at the pachytene stage (during fetal development) quite clearly demonstrate that failure of synapsis and chiasma formation is much more common than at the corresponding stage in males.

Failure of crossover/chiasma formation means that maternal and paternal homologous chromosomes appear as disoriented univalents, leading to random rather than regular segregation. Daughter cells may thus receive a paternal, a maternal, both or none of these (non-exchanged) chromosomes. One other complication of this situation is the potential for chromatids of univalent chromosomes to undergo precocious separation followed by disoriented segregation, leading to daughter cells receiving one, two or no chromatids, instead of a whole (maternal, paternal or recombinant maternal-paternal chromosome). Segregation errors, including those of chiasmate bivalents (see Fig. 6) are particularly common in human females, underlying the high rate of aneuploid offspring in our own species.

Figure 6: Single-chiasma bivalent showing normal disjunction at MIAI, where kinetochores are oriented towards opposite spindle poles (left); traditional non-disjunction at AI where homologus kinetochores are orientated towards the same pole (middle); Precocious sister chromatid separation and MI-AI transition leading to respectively an extra and missing chromatid in segregants (right).

Numbers of additional chiasmata are dependent on chromosome length

It is clear from observation on human spermatocytes that numbers of additional chiasmata over and above the obligate are dependent on chromosome length. Successive chiasmata are separated by large (many Mb) chromosome segments, a phenomenon termed chiasma interference (Fig 7). The mechanism behind chiasma interference is unknown.

Figure 7: The distribution patterns of MLH1 foci on chromosome 21 SCs in oocytes illustrating the difference with one in comparison to two foci, which are widely spaced apart.

Measurements of distances between MLH1 crossover foci on the SCs at pachytene indicates interference distances in terms of physical SC length to be basically the same in human spermatocytes and oocytes. This may ‘explain’ the higher  chiasma  frequency in  human females in comparison to males because the SCs in oocytes are relatively ‘decondensed’ and much longer than in spermatocytes. Chiasmata also have some tendency to sex-specific preferential positioning with a stronger accumulation near ends of chromosomes in males than females (see Fig 8).

Figure 8: Genetic maps of chromosome 21, based on MLH1/chiasma data, illustrating high amount of distal crossovers and the corresponding expansion of genetic length distally in the male (left) in comparison to that in the female (right).

Chiasma maps may be used for the construction of genetic maps

Chiasma maps allow us to distinguish between reciprocal recombination (crossing-over/chiasma formation) and other types of recombination events (such as sister chromatid exchange and conversion). In addition, the direct meiotic crossover analysis allows estimates of genetic map distances as well as recombination fractions from the same observed raw data.

Haldane (1919) originally defined the Morgan Unit of genetic map distance as that length of a chromatid that has experienced on average one crossover per chromatid. Each chromatid in this situation corresponds to a potential gamete (Fig 2). As each crossover event (MLH1 focus/chiasma) may give rise to two recombinant as well as two non-recombinant gametes, then the genetic map distance (Morgan, M) is calculated as half the average number of MLH1 foci or chiasmata in the interval concerned. In other words the genetic map distance in centimorgans (cM) is obtained by multiplying the average number by 50, thus an average of 50 autosomal chiasmata in spermatocytes corresponds to a male genetic map length of 2,500 cM, while 70 MLH1 foci in oocytes translate to a female genetic map length of 3,500 cM.

The construction of genetic maps from chiasma/crossover/recombination data obtained by cytogenetic analysis in individual cells has many advantages. Above all this direct approach allows sex-specific estimates of both intra- (intercellular) and inter-individual variation in patterns of meiotic recombination with respect to the whole genome, individual chromosomes and chromosome segments.

The most recent linkage based genetic maps agree with the chiasma-based

Linkage analysis uses statistical methods to infer the likely crossover pattern by studying the co-segregation of alleles with nearby DNA markers between parents and their children. From this the order of DNA markers is determined together with estimates of recombination frequency between the markers.

Earlier linkage-based male-specific genetic maps showed large discrepancies in relation to those based on cytogenetic data with a general tendency for inflation in the length of individual chromosomes (likely due to overcompensation for presumed double recombinants). However, the most recent linkage-based genetic maps agree with the chiasma based maps, at least as regards total genetic map length for each individual chromosome.

Chromosome segregation efficacy can be estimated by MII analysis

From Pachytene until MI, all four chromatids are ‘glued’ together along their entire length (Fig 1). At MI each pair of sister chromatids are present as a unit with both sister kinetochores facing in the same direction, mono-orientation as mentioned earlier. This means that both sister chromatids of a homolog are pulled to the same pole, once AI is initiated. Once initiated, this is seemingly a very rapid process, as indicated by, for example, the absence of any AI spermatocytes in preparations from testicular biopsy samples. At MII the sister chromatids are bi-oriented, as in mitosis, to allow segregation of one chromatid per daughter cell during AII.

A wealth of information about the complex interaction of proteins between kinetochores and spindle microtubules at the MI to AI transition has recently accumulated. However, there are yet no direct microscopy studies highlighting this very important process in humans. On the other hand, evaluation of the efficacy of AI segregation can be made by chromosome analysis of cells at the MII stage.

MII analysis indicates male AI segregation errors are rare

Information in the male on AI segregation may be obtained by investigation of spermatocytes at MII, prepared from testicular biopsies in a way very similar to that used for making preparations from blood lymphocytes for somatic chromosome analysis. One particular problem here is that chromatids of individual MII chromosomes in spermatocytes prepared this way often splay to the extreme and chromatids are often slightly separated. In addition MII chromosomes are loosely coiled and have a tendency to hook into each other (Fig 5). This makes even counting of chromosomes problematical.

We have used stringent criteria to avoid artefacts and did then not detect any numerical abnormalities in 200 MII spermatocytes from 6 normal men with apparently normal mitotic karyotypes. Thus, there was no indication of either extra or missing chromosomes or indeed any extra or missing chromatids. To our knowledge no other studies have to date been presented on this issue. We may tentatively conclude that AI mal-segregation in the human male is rare with the anueploidy rate in spermatocyte MIIs being less than 0.5-1%.

It may be added that chromosome analysis of mature sperm shows that around 2-3% of mature spermatozoa are aneuploid. Investigations of individual chromosomes indicate that average numerical abnormality rate for each is around 0.1-0.2%. Only chromosomes 21 and 22 and the XY pair show slightly increased rates of sperm aneuploidy. The question whether or not some men may be predisposed to aneuploid offspring has not yet been answered with any certainty. On the other hand, tracing of DNA markers between fathers and XXY Klinefelter sons has demonstrated reduced XY recombination, where the implication is that this has led to XY disomy in sperm. It is also important to note that FISH analysis of sperm from carriers of structural chromosome rearrangements, such as translocations, has shown a high rate of imbalance for the chromosomes involved.

MII analysis indicates female segregation errors are very common

The apparently rare occurrence of male segregation errors stands in sharp contrast to that in the human female. Chromosome analysis of a large population of oocytes at the MII stage has been performed. These studies demonstrate quite clearly that oocyte segregation errors involving both whole chromosomes and chromatids are very common indeed. It should be noted, however, that the material investigated so far consists almost exclusively of MII oocytes obtained at infertility (IVF) treatment, where spare oocytes, spontaneously arrested at MII, have been donated for research and used for karyotyping.

The morphology of oocyte MII chromosomes makes them slightly more amenable to karyotyping than is the case for spermatocytes; chromatids do not normally splay to the extent that they do in spermatocytes. Oocyte MII chromosomes are also more condensed and therefore more easily spread and separated from each other. The largest study, using traditional chromosome banding (R-banding), concerned 1,397 MII oocytes. Chromosomal abnormalities were detected in around one fifth (20.1%) of oocytes analysed. Aneuploidy, of chromosomes or chromatids, was detected in 10.8% of the cells, while structural abnormalities (breaks, deletions and acentric fragments) were much more rare, seen in 2.1% of cells. More recently, a total number of 14 studies, using multicolour FISH, applying several rounds of probes consecutively, has yielded quite variable results with MII aneuploidy rates ranging from 3.0-45.5 %. The most stringent protocols involve the combination of centromeric or locus-specific probes together with whole chromosome paints, allowing precise identification of both whole chromosomes and chromatids.

The consensus from all these studies, including the FISH (and other molecular) investigations and involving both MII occytes and polar bodies, is that around 15-20% of human oocytes are aneuploid as a result of both traditional non-disjunction and premature chromatid separation. AI segregation errors seem to be more common that those at AII and more commonly affect small chromosomes than the larger ones. Most importantly, the rate of single chromatid non-disjunction has been found to be positively correlated with age of the oocyte donor. Taking all the cytogenetic evidence together, it therefore seems likely that the maternal age effect of common aneuploidies, in particular Trisomy 21 Down Syndrome, is caused by a combination of problems in chiasma formation during fetal life and cohesion of centromeres/kinetochores at AI, taking place just before ovulation.

There are no direct studies highlighting the meiotic cause of genetic disease

Hardly anything is known so far from direct meiotic studies on the mechanisms of origin of structural chromosome rearrangements, including for example, extra marker chromosomes, Robertsonian and reciprocal translocations, inversions, insertions, deletions and duplications. Some of these may arise from breakage and repair processes taking place either pre- or post- meiosis.

Special attention has been paid to the origin of de novo deletions and duplications, comprising so-called genomic disease. The suggestion from somatic DNA marker investigations of children and their parents is that these chromosome disorders originate by misalignment of homologs at Prophase I, followed by ‘ectopic’ crossing-over/chiasma formation/recombination between similar (paralogous) DNA sequences, located at distance from each other, either within the same homolog or indeed on different homologs. To our knowledge no investigations have yet been performed to identify such ectopic crossovers directly by investigation of SCs. However, misalignment has been clearly documented in human oocytes, where centromere signals (by the CREST antibody) are often staggered.

It is known from previous indirect studies, tracking DNA polymorphisms between parents and children, that patterns of meiotic recombination (deciding the formation of bivalents and their shape) play an important role for the origin of common aneuploidies such as Trisomy 21 Down Syndrome. However, it is not yet clear how this may work in detail. Most segregational errors underlying these aneuploidies have been found to take place at maternal AI; lack of chisma formation is one obvious reason. It has also been suggested that certain chiasma positions, such as a single distal chiasma of bivalent 21 may constitute a specific risk factor. On the other hand, the very same position is one of the most common in spermatocytes, where AI segregation errors are by comparison very rare. Clearly a so-called ‘second hit’ must be of paramount importance in handling such oocyte bivalents differently from spermatocytes at a later stage. What this would be is yet unclear but it has been proposed that age-related increased risk for Trisomy 21 might be related to deterioration in the cohesion proteins, holding chromatids and centromeres together during Meiosis I. Remarkably, there are also clear indications that Meiosis I spindles in oocytes obtained from older women are much more irregular than those from younger women. More detailed information on the intricate interaction between chromosomal and spindle proteins in oocytes at MI and AI in particular may provide some answers including an explanation for the well-known maternal age effect. Many factors may contribute to the formation of spindles including a range of spindle proteins, mitochondrial and hormonal status, follicle maturation in relation to the oocyte pool and peri-follicular microcirculation, and age-related changes in any of these factors may play a role.

Acknowledgement: This review is a revised version of a paper by Hultén, M., Baker, H. and Tankimanova, M. (2005) Meiosis and meiotic errors. In Encyclopedia of Genetics, Genomics, Proteomics and Bioinformatics (online edition), Jorde, L.. B., Little, P. F. R.., Dunn, M. J. and Subramaniam, S. (Eds) John Wiley & Sons Ltd: Chichester. DOI: 10.1002/047001153X.g102206.

(Copyright John Wiley & Sons Limited. The permission to reproduce the figures is greatfully acknowledged.)

Address of the author:

Maj Hultén
Department of Biological Sciences
University of Warwick
COVENTRY CV4 7AL
UK
E-mail: maj.hulten@warwick.ac.uk